Human sensory systems are continuously bombarded with far more input than can be processed at a single time. As a result, attentional mechanisms have evolved so that available processing capacity is dedicated to encoding only the most salient and behaviorally relevant stimuli in the environment. In turn, the most important stimuli dominate perceptual awareness and have privileged access to the neural mechanisms that control decisions about how to best interact with external objects. Recently, investigators have proposed the counterintuitive hypothesis that selective attention operates to gate relevant sensory information in time with slow intrinsic cortical oscillations in the theta and alpha bands (i.e. ~4-12Hz). For example, behavioral performance on basic visual tasks waxes and wanes in time with alpha oscillations, and experimentally disrupting these rhythms impairs perception. Here we will test the hypothesis that these rhythmic fluctuations in behavior reflect oscillations n the fidelity of sensory representations in early cortical areas (sensory-modulation hypothesis). We will use a combination of formal models of decision making coupled with novel EEG analysis techniques that can non- invasively measure feature-selective representations with a high degree of temporal resolution. Thus, this work will test a focused hypothesis about the nature of intrinsic cortical oscillations and their link to rhythmic changes in human information processing. In addition, our novel approach to analyzing high-dimensional EEG data sets will provide a non-invasive and well-tolerated means of measuring feature selective responses in human cortex with high temporal resolution, in line with the strategic aim of the NIH to develop novel tools and methodologies for understanding how populations of neural cells work together within and between brain regions.